A newly credentialed gastroenterologist has just finished her fellowship and joined a practice four states away from where she trained, because that is where the job was.
Her clinical skills are exactly where they should be. She has done the case volume, passed her boards, and can walk into a procedure room as competent as anyone in the building. What she does not have is a single referring physician in this market who has ever sent her a patient, because none of them have ever worked with her, watched her manage a complication, or heard from a trusted colleague that she is good.
Her calendar for her first month has eleven patients on it. The gastroenterologist two offices down, eleven years into practice in this same market, has a calendar booked six weeks out. The difference between them is not measurable in any board exam. It is measurable only in a referral graph that one of them spent a decade building and the other has not yet had a single day to start.
Meanwhile, roughly two hundred miles away, three of her former co-fellows are already established, seeing patients, and would send her a referral without hesitation if a patient of theirs needed a gastroenterologist in her new city. None of them know she has moved there. The trust that would fill her calendar already exists. It is simply dormant, and nothing in the system that recruited her is built to wake it up.
The productivity gap is real, large, and driven by neither skill nor effort
Start with the direct evidence, because this is not an impression new physicians share informally. It has been measured precisely.
A national cohort study of 34,878 Veterans Health Administration attending physicians, published in the Journal of General Internal Medicine in 2025, found that new hires saw 1.72 fewer patient encounters per clinic day than tenured colleagues in their first quarter, a statistically tight finding with a 95 percent confidence interval of negative 1.79 to negative 1.65. That gap did not close quickly. It narrowed to just 0.44 fewer encounters per day by the eighth quarter, meaning it took roughly two full years for a new attending physician to reach something close to typical productivity.
The population studied here is important context: these are board-certified attending physicians, already credentialed, already clinically competent, joining a system where referral and patient-flow relationships have to be rebuilt regardless of how skilled the physician is. The gap is not explained by anything the new physician does not yet know how to do clinically. It is explained by a graph, of referring colleagues, of scheduling familiarity, of institutional trust, that a tenured physician has and a new one does not, and that takes about two years to rebuild from nothing.
The dollar cost lands entirely on the practice that recruited her
This gap is not merely a professional inconvenience for the new physician. It is a specific, quantifiable financial exposure carried almost entirely by whoever hired her.
Physicians generate, on average, roughly $2.4 million a year in revenue once fully productive, according to AMN Healthcare's physician-revenue survey data. Applying the VA study's productivity curve, roughly 25 percent below steady-state output in the first quarter, closing to roughly 6 percent below by the eighth quarter, to a new specialist earning something in the range of half that average figure during her two-year ramp implies on the order of $200,000 to $400,000 in foregone practice revenue per new hire, an estimate this article is deriving from those two data points rather than quoting from a single audited source, and it should be read as an order-of-magnitude illustration rather than a precise figure.
That cost is layered on top of what the practice already spent to recruit her in the first place. The median time to fill a physician search is 118 days, and offer-acceptance rates have been falling, from 83 percent to 71 percent year over year, according to the American Association for Physician Recruitment's 2025 benchmarking report, indicating both the search itself and the risk that a found candidate declines are both rising. Add a guaranteed salary period, standard in most new-physician contracts, and a recruiting practice is paying full compensation against a productivity curve it did not build and has no established way to shorten.
Scale this across the roughly 40,000 residency and fellowship graduates entering practice each year, and even a conservative assumption that a meaningfully faster ramp is achievable for a subset of them implies a national opportunity measured in the hundreds of millions of dollars in accelerated productivity.
The dormant asset: her own training cohort
Here is the finding that reframes this from an unavoidable cost of being new into a specific, addressable coordination failure.
This series has previously documented that co-training ties, having trained in the same residency or fellowship program with overlapping years, are the single strongest measured predictor of referral trust in American medicine. Co-trained specialists received 26.2 percent of eligible PCP referrals versus a 21.4 percent baseline, a statistically significant gap, and patients seen by a co-trained specialist rated the visit 9.0 percentage points higher, described by the researchers as the equivalent of moving from the median to the 91st percentile in patient experience, per a 2023 study in JAMA Internal Medicine building on the original 2021 Health Services Research finding.
And yet, of the referrals where an eligible co-trainee existed, only 8.0 percent actually went to one. The trust tie that produces measurably better outcomes and measurably higher referral rates when activated is, overwhelmingly, sitting unused, because nobody tracks where a physician's former co-residents and co-fellows currently practice, and nobody connects that information to the moment it would matter most: the day a new specialist opens her doors in an unfamiliar market.
A new gastroenterologist landing in a city where two or three of her former co-fellows already practice is not, in fact, a stranger to that market's referral graph. She is one warm introduction away from a functioning piece of it. The introduction simply never happens, because the information needed to make it, who trained with whom, and where they are now, exists only in scattered individual memories.
Why the current substitutes do not touch this
Every mechanism a new specialist or her recruiting practice currently uses to build referral volume addresses strangers, not the cohort ties that would work fastest.
Physician-liaison outreach, the standard tool practices use to build referral volume, sends generic marketing to every referring practice within a radius, regardless of whether any prior relationship exists. It is undifferentiated outreach to strangers, competing for attention against every other specialist doing the same thing in the same market.
Residency and fellowship alumni offices treat their graduates as a donor list, not a referral-activation network. The infrastructure to know where alumni currently practice exists in most programs in some form, for fundraising purposes; it has essentially never been repurposed to tell a newly arrived graduate which of her own cohort is already established nearby.
Doximity holds residency-program fields on physician profiles and could, in principle, surface this information. It has never activated it as a referral-routing tool, and it has no reciprocity mechanism that would make a co-trainee obligated, or even prompted, to actually make the introduction once identified.
Health system referral analytics platforms like Definitive Healthcare and Trilliant look backward at existing referral patterns to identify leakage and steering opportunities. They have no forward-looking function for a new hire who has no referral pattern yet to analyze.
Every one of these tools is built to solve a different problem than the one a new specialist actually has. None of them asks the single most useful question available at the moment of her launch: who did you train with, and where are they now?
Why this is getting more expensive, not less
Three forces are compounding the cost of an unactivated cohort tie at exactly the moment it would be cheapest to fix.
Physician employment has passed 75 percent, meaning most new specialists are launching cold into an employer-selected market rather than building an organic practice over a decade in a place they chose and where they may already have connections. The cold start used to be a choice some physicians made; it is now close to the default condition of entering practice.
Recruitment and vacancy costs are rising in parallel. With a 118-day median search time and acceptance rates falling toward 71 percent, every month a newly placed specialist spends ramping up is a month layered on top of an already expensive and increasingly uncertain recruiting process.
The evidence base for the fix has only recently become solid enough to act on. The co-training referral-quality finding is now published and replicated across two studies, giving a launch-acceleration approach an evidence base it simply did not have a decade ago, when the co-training effect was closer to folk wisdom among physicians than a measured, citable finding.
Why nobody owns the activation
This is the structural diagnosis. The dormant asset is real, the evidence for its value is published, and still nobody has built the mechanism to activate it.
Recruiting practices and health systems have no visibility into a new hire's training cohort's current locations. This information was never collected as part of credentialing or onboarding, because nobody framed it as relevant to referral volume until the co-training research existed to justify collecting it.
Residency and fellowship programs treat alumni as a fundraising list, not a referral-activation network, for the same reason liaison firms sell generic outreach: nobody has built the connective layer between "we know where our graduates practice" and "we could introduce a newly arrived graduate to the ones nearby."
Physician-liaison vendors sell what they know how to sell: outreach to every practice in a radius. A targeted product built around a specific new specialist's verified prior colleagues would require access to a co-training graph that, as this series has documented elsewhere, essentially does not exist anywhere in queryable form.
The result is a genuinely valuable asset, a new specialist's own dormant trust relationships, sitting unused at precisely the moment activating it would be most valuable, because every party positioned to activate it has built its business around a different, less effective mechanism instead.
What would actually work
Cross-reference training history against target market at credentialing, not after. The moment a new specialist accepts a position is the moment her training cohort's current locations should be checked against her new practice's referral radius, not discovered informally months into a slow ramp.
Facilitate the introduction, not just the identification. Knowing that three former co-fellows practice nearby accomplishes little without an actual warm handoff, a structured introduction that makes clear both that she is now practicing locally and that she is accepting referrals.
Build this as a day-one product, not a six-month follow-up. The VA data shows the productivity gap is steepest in the first quarter; a launch-graph activation that happens on day ninety has already missed the period where it would do the most good.
Track outcomes back to the co-training research, not just anecdote. A member-reported measure of time-to-steady-state referral volume, segmented by whether a verified cohort-activation handoff occurred, would let the field actually test whether activating dormant trust compresses the VA study's two-year curve, rather than assuming it does.
Keep it legally clean by design. Any structured warm-introduction mechanism between a new specialist and potential referrers needs to be built with Stark Law and Anti-Kickback Statute considerations in mind from the outset, an introduction based on a genuine prior professional relationship is a fundamentally different thing from a paid referral arrangement, and the two need to stay visibly distinct.
Make it easy for the established co-trainee to say yes. The 8.0 percent activation rate on eligible co-trainee referrals is not evidence that established physicians are reluctant to refer to former colleagues; it is far more likely evidence that they have no efficient way to know a former colleague has arrived nearby and is accepting patients. Lowering the effort required to make that introduction is likely to matter more than any incentive.
What you can do now
If you are a newly practicing specialist
Find out where your former co-residents and co-fellows are practicing before you finish your first month. This is a concrete, doable task, a handful of messages or a search through your own training program's alumni contacts, and the evidence suggests it is the single highest-leverage referral-building action available to you.
Reach out directly and specifically. Tell them where you have landed and that you are accepting referrals. The published research suggests co-trained referral relationships work when activated; the missing step in most cases is simply that nobody made the ask explicit.
Do not assume liaison marketing will substitute for this. Generic outreach to unfamiliar referring practices in your area is a reasonable complement, but it is competing for attention against every other new specialist doing the same thing, while a former co-fellow already trusts your clinical judgment and needs only to know you are there.
Expect the ramp, and plan your finances around roughly two years, not two months. The VA data's productivity curve is a genuine, measured pattern, not a personal failing, and knowing the shape of it in advance should change how you and your practice structure a guaranteed-salary period.
If you recruit or lead a practice hiring a new specialist
Ask about training history as part of onboarding, and use it. A newly hired specialist's residency and fellowship programs, and the years she trained, are already collected for credentialing purposes; cross-referencing that information against where her cohort currently practices costs almost nothing and could meaningfully shorten the productivity ramp you are already paying for.
Reconsider your liaison marketing budget in light of this. Generic physician-liaison outreach and a targeted cohort-tie introduction are not mutually exclusive, but if you are spending on the former and ignoring the latter, you are likely spending more to reach a colder audience than the warmest one available to your new hire.
Structure guaranteed-salary periods with the VA curve in mind. A roughly two-year path to steady-state productivity is a measured, replicable finding, not a worst case, and a compensation structure built around a shorter assumption is setting both the practice and the new physician up for a mismatch neither one caused.
If you build referral, credentialing, or physician-recruitment software
Build the launch-graph product nobody else has. Every existing referral-analytics tool looks backward at established patterns. A tool that identifies a new specialist's verified training-cohort ties against her new market at the moment of hire is solving a problem no current vendor addresses, using data (training program, years, current practice location) that is largely already collected for other purposes and simply never cross-referenced.
Design the consent and Stark-safe handoff mechanics from the start. A credible version of this product needs a clear boundary between a warm professional introduction based on genuine training history and anything that could be construed as a paid referral arrangement, built into the product rather than added after a legal review flags it.
Frequently asked questions
How long does it take a new physician to reach full patient volume? Roughly two years, based on the best available evidence. A national VA cohort study of 34,878 attending physicians found new hires saw 1.72 fewer patient encounters per clinic day than tenured colleagues in their first quarter, narrowing to 0.44 fewer encounters by the eighth quarter, published in the Journal of General Internal Medicine (2025).
Why don't new specialists get many referrals right away? Because referral trust is built through direct relationships and observed outcomes over time, and a newly arrived specialist has not yet had the chance to build any, regardless of her clinical skill. This is a referral-graph and patient-flow problem, not a competence gap; the physicians studied in the VA cohort were already board-certified attendings.
How much does the productivity ramp cost a practice that hires a new specialist? There is no single precise industry figure, but combining AMN Healthcare's estimate of roughly $2.4 million in average annual physician-generated revenue at full productivity with the VA study's productivity curve suggests a plausible range of roughly $200,000 to $400,000 in foregone revenue per new hire during the ramp period, an order-of-magnitude estimate rather than an audited figure.
Do doctors get more referrals from people they trained with? Yes, measurably. Co-trained specialists received 26.2 percent of eligible PCP referrals versus a 21.4 percent baseline, and patients seen by a co-trained specialist rated the visit 9.0 percentage points higher, described as moving from the median to the 91st percentile, according to 2021 and 2023 studies in Health Services Research and JAMA Internal Medicine. Only 8.0 percent of eligible co-trainee referral opportunities are actually used.
How do new doctors build a referral network? Currently, mostly through generic physician-liaison marketing to unfamiliar practices in the area, direct outreach to any personal contacts they happen to maintain, and slow, organic relationship-building over months or years. There is no established mechanism that systematically connects a new specialist to her own former training cohort at the moment she launches in a new market, despite evidence that those ties are unusually effective when activated.
How long does it typically take to fill a physician search, and how does that affect ramp costs? The median time to fill a physician search is 118 days, with offer-acceptance rates falling from 83 percent to 71 percent year over year, according to the American Association for Physician Recruitment's 2025 benchmarking report. That rising search cost and risk is layered on top of the roughly two-year productivity ramp once a physician is placed, compounding the total cost of each hire.
The bottom line
A new specialist's first two years are not slow because she is still learning to practice medicine. The VA data is unambiguous on this point: these are already board-certified attending physicians, and the productivity gap, 1.72 fewer encounters a day at the start, narrowing over roughly two years, tracks a referral and patient-flow graph rebuilding itself from nothing, not a skill deficit closing.
That two-year rebuild is largely unnecessary in its current form, because a meaningful piece of the graph she needs already exists. Her own former co-residents and co-fellows, scattered across the country, are the single strongest measured predictor of referral trust in the published literature, worth a documented 9-point jump in patient experience ratings when activated. And in the overwhelming majority of cases, that tie sits completely dormant, because nobody tracks where a physician's training cohort currently practices and nobody connects that information to the day she needs it most.
Every existing tool, liaison marketing, alumni fundraising, referral analytics, is built to solve a different problem. None of them asks the one question that would actually shorten her ramp: who did you train with, and where are they now?
That gastroenterologist is going to spend her first year building a referral base from scratch in a city where three people who already trust her clinical judgment happen to practice, and none of them know she has arrived. The trust is not missing. It is simply asleep, and nobody has built the alarm clock.
Part of a series on the missing professional infrastructure of healthcare. Previously: Who Gets Asked
Evidence note: the new-physician productivity curve (1.72 fewer encounters per day at quarter one, narrowing to 0.44 by quarter eight, across 34,878 VHA attending physicians) is from Li, Barr, Pizer, Garrido, and Thorsness, Journal of General Internal Medicine (2025). Physician revenue and vacancy-cost figures are from AMN Healthcare's physician-revenue survey data; the resulting $200,000 to $400,000 per-hire cost estimate is this article's own illustrative calculation combining those two sources and should be read as directional, not as an audited figure. Co-training referral figures (26.2 percent versus 21.4 percent baseline, 8.0 percent activation rate, 9.0-point patient-rating gap) are from Pany et al., Health Services Research (2021), and a related 2023 study in JAMA Internal Medicine. Physician search and acceptance-rate figures are from the American Association for Physician Recruitment's 2025 benchmarking report. The VA cohort, while a large and rigorous national dataset, is drawn from a single-payer, salaried system, and the productivity curve's generalizability to fee-for-service private practice, where referral competition works differently, has not been independently confirmed in the sources reviewed for this article. The opening scenario is a composite illustration built from the patterns documented in this evidence, not a specific reported case.
